In recent years, the amalgam of deep learning fundamentals with Natural Language Processing techniques has shown a great improvement in the information mining tasks on unstructured text data. The models are now able to recognize natural language and speech comparable to human levels. Despite such improvements, discrepancies in the results still exist as sometimes the information is coded very deep in the syntaxes and syntactic structures of the corpus. User: Hi, I took a horrible picture in a museum, can you tell where is it located? User: Hi, I took a horrible picture in a museum, can you tell where is it located?

Google announced today it will open a lab in Beijing dedicated to researching artificial intelligence (AI). The news comes as China's government and tech companies race ahead to dominate the field, putting Google in a position where it has no choice but to set up locally in order to remain at the cutting edge. According to the search giant, the lab will be led by a small team of researchers and supported by Google engineers already working in China on developing the company's global-facing products. It will be helmed by Jia Li, head of research and development at Google Cloud AI, and Fei-fei Li, director of the Stanford Artificial Intelligence Lab and a chief scientist at Google Cloud AI. In a blog post announcing the move, Li notes that the company is currently hiring for positions at the lab--something media outlets picked up on a few months ago. Searching through Google's job-posting page for Beijing yields recruitment ads for technical leads and software engineers specializing in machine learning.

"Will China Win The Artificial Intelligence Race?" is part of Aviation Week & Space Technology's subscription package. Subscribe now to read this full article. And by subscribing, you'll also receive full coverage of what's next in technology from the experts trusted by the global aerospace & defense community.

As you may probably know, DeepMind has recently published a paper on AlphaZero [1], a system that learns by itself and is able to master games like chess or Shogi. Before getting into details, let me introduce myself. I am a researcher in the broad field of Artificial Intelligence (AI), specialized in Natural Language Processing. I am also a chess International Master, currently the top player in South Korea although practically inactive for the last few years due to my full-time research position. Given my background I have tried to build a reasoned opinion on the subject as constructive as I could.

Near isometric orthogonal embeddings to lower dimensions are a fundamental tool in data science and machine learning. In this paper, we present the construction of such embeddings that minimizes the maximum distortion for a given set of points. We formulate the problem as a non convex constrained optimization problem. We first construct a primal relaxation and then use the theory of Lagrange duality to create dual relaxation. We also suggest a polynomial time algorithm based on the theory of convex optimization to solve the dual relaxation provably. We provide a theoretical upper bound on the approximation guarantees for our algorithm, which depends only on the spectral properties of the dataset. We experimentally demonstrate the superiority of our algorithm compared to baselines in terms of the scalability and the ability to achieve lower distortion.

The goal of counterfactual learning for statistical machine translation (SMT) is to optimize a target SMT system from logged data that consist of user feedback to translations that were predicted by another, historic SMT system. A challenge arises by the fact that risk-averse commercial SMT systems deterministically log the most probable translation. The lack of sufficient exploration of the SMT output space seemingly contradicts the theoretical requirements for counterfactual learning. We show that counterfactual learning from deterministic bandit logs is possible nevertheless by smoothing out deterministic components in learning. This can be achieved by additive and multiplicative control variates that avoid degenerate behavior in empirical risk minimization. Our simulation experiments show improvements of up to 2 BLEU points by counterfactual learning from deterministic bandit feedback.

A number of the most popular websites and services online, including Facebook and PayPal, are vulnerable to an exploit which has resurfaced from 1998. The security flaw, dubbed ROBOT, was first discovered almost two decades ago by Daniel Bleichenbacher. PKCS #1 1.5 padding error messages produced by secure sockets layer (SSL) servers allow for an adaptive-chosen ciphertext attack which "fully breaks the confidentiality of TLS when used with RSA encryption," according to researchers Hanno Böck and Juraj Somorovsky from Hackmanit GmbH, Ruhr-Universität Bochum, and Tripwire VERT's Craig Young. The server implementation bug could be used to perform RSA decryption and key signing in order to decrypt traffic. "We discovered that by using some slight variations this vulnerability can still be used against many HTTPS hosts in today's Internet," the team says.

I had the pleasure of meeting Shing Tao, the CEO of Remark Holdings (NASDAQ:MARK), on Monday night. He had just flown into San Francisco from Beijing and I was meant to be his first meeting before dinner and some shut eye; he had a big Tuesday in front of him with Roth Capital (whose analyst loves MARK, although they haven't been their bankers) taking him on a Non-Deal Roadshow and introducing him to several institutional investors. Remark has been very hot lately, jumping almost 200% in value since the company reported Q3 earnings in mid-November. The move came on the back of 2018 revenue guidance for their Artificial Intelligence operating subsidiary, KanKan. They guided to $30M in KanKan AI revenues for next year, which would be a 500% increase over 2017.

It's a tradition that's been around since the 1936 Olympic Games in Berlin, but now the torch relay has been given a very modern update. During the 41st day of the relay in the lead-up to the Pyeongchang Winter Olympics in South Korea, a humanoid robot carried the torch. The robot, called Hubo, walked about 150 metres (500 feet) to a wooden wall, before using a drill to cut a hole and pass the torch through. The robot participated in the relay past the Korean Advanced Institute of Science and Technology (KAIST), where it was created. A video from KBS News shows the incredible moment in action.

You have worked for weeks on building your machine learning system and the performance is not something you are satisfied with. You think of multiple ways to improve your algorithm's performance, viz, collect more data, add more hidden units, add more layers, change the network architecture, change the basic algorithm etc. But which one of these will give the best improvement on your system? You can either try them all, invest a lot of time and find out what works for you. You can use the following tips from Ng's experience.